TORONTO — For notebook users who also happen to be frequent flyers, elbow-room is a big deal.
That’s why IBM made sure its Thinkpad places the keyboard as close to the screen as possible. The machine also features a tiny light attached to the top of the screen so users can type on a plane without
having to turn on the overhead seat light.
Little product improvements like that are the result of user-focused research development at Big Blue. The company opened a User-Centred Design (UCD) lab in Toronto about two years ago. Using a combination of market research, online surveys and in-depth, on-site user studies, the lab helps to oversee and refine all of IBM’s product releases from DB2 to Tivoli to WebSphere. In the case of Thinkpad development, IBM brought actual plane seats into the lab and developed the product with those ergonomics in mind.
“”It’s really taking methods from science and a lot of other disciplines . . . baked into applied research,”” said Karel Vredenburg, program director for the UCD lab.
According to Vredenburg, design (or lack of it) is responsible for 75 to 80 per cent of user help desk calls rather than a software or mechanical flaw in the product itself.
There are several essential steps built into the UCD approach to products: Who will use it; what those users are looking for; what similar products are available from the competition; coming up with an initial product design; improving it based on feedback; and running benchmark tests on a completed product.
Test users undergo sessions that typically take two to four hours — but they’re treated to a meal first to make them feel more like guests than guinea pigs. They may be asked to comment on a “”low fidelity prototype,”” which could simply be a prospective Web site design roughed out on a piece of paper. “”The user doesn’t feel like anything is set in stone,”” said UCD team member Mike Wulkan. Users are comfortable marking up a paper design and making their own suggestions.
A “”high fidelity prototype”” could be that same Web design but actually coded and displayed on a monitor. Two users test the design — one operates the mouse, the other the keyboard. Separating the tasks encourages communication and discussion between the participants, said Wulkan.
Product developers watch the test session behind one-way glass and can hear the user comments as they are made. Wulkan said that he’s seen developers cringe as they witness this process — design elements that seem simple and intuitive to them may befuddle some users. “”This is an extraordinarily fun way of illustrating to them that they might have to step out of their own boxes,”” said Wulkan.
IBM consulted its UCD division for a technology that’s actually designed to remove the user from the experience. Autonomics, sometimes referred to as “”self-healing,”” is a way to allow products like DB2 to handle low-level maintenance tasks without human intervention.
But it’s a difficult balance to strike — how many tasks can IBM realistically take out of the hands of database administrators and code into the database itself? Part of the design team’s challenge was figuring out how much the system should do and how much the user should do, said Vredenburg.
“”You don’t want them spending a whole lot of time focusing on the infrastructure . . . automomics UCD is really at the heart of determining those things,”” he said. On the other hand, “”you can just say, ‘OK, let’s automate everything. Let’s turn this autonomic thing on and if it crashes, I lose my job.'””
DB2 8.1 was the first database product that introduced autonomic elements, specifically a “”health monitor”” to oversee and correct operations. “”This gets down to a trust issue . . . Is this person going to be able to trust what this thing is automating for them?”” said Rick Sobiesiak, IBM Canada’s manager for DB2 user-centred design.
Stephen O’Connell, another member of the design team, added that IBM was able to set limits on autonomic capabilities based on information gleaned from DBAs by asking them to outline typical database scenarios and problems.
For example, with the DB2 health monitor, no actions are required for routine maintenance tasks, but a human steps in when a decision has to be made that will affect the overall wellbeing of the database.
The research and testing that the UCD lab conducts is a useful guidepost for IBM brass, but they’re not always wedded to the results. Sometimes competitive pressures may urge IBM to push a product into the market early, said Vrendenburg, but then the company has to be prepared for negative feedback from the trade magazines that review the product and the early adopters that buy it.
Research is a useful tool, he added, but manufacturers can always expect to make tough decisions on the best time to enter the market.
— Illustration by Jarrett Osborne